Make IC patching resilient to flushing of the original target() ic.

Make IC patching resilient to flushing of the original target() ic.

Committed: https://code.google.com/p/v8/source/detail?r=13831

[Toon Verwaest, 2013-03-05 17:10:56]

PTAL

[Jakob Kummerow, 2013-03-05 17:32:59]

LGTM with a real comment and a nit.

https://chromiumcodereview.appspot.com/12451003/diff/8016/src/ic.cc
File src/ic.cc (right):

https://chromiumcodereview.appspot.com/12451003/diff/8016/src/ic.cc#newcode1041
src/ic.cc:1041: AssertNoAllocation no_gc;
This AssertNoAllocation (and its scope) seem unnecessary, since you have one on
line 1035 too.

https://chromiumcodereview.appspot.com/12451003/diff/8016/src/stub-cache.cc
File src/stub-cache.cc (right):

https://chromiumcodereview.appspot.com/12451003/diff/8016/src/stub-cache.cc#n...
src/stub-cache.cc:907: Code::StubType type = handlers->length() == 1
nit: I have a weak preference for the following formatting, but you can also
keep what you have if you prefer:

  Code::StubType type = handlers->length() == 1 ? handlers->at(0)->type()
                                                : Code::NORMAL;

[Toon Verwaest, 2013-03-05 17:37:14]

Addressed comments, landing.

https://chromiumcodereview.appspot.com/12451003/diff/8016/src/ic.cc
File src/ic.cc (right):

https://chromiumcodereview.appspot.com/12451003/diff/8016/src/ic.cc#newcode1041
src/ic.cc:1041: AssertNoAllocation no_gc;
I removed the one above. The idea was to move that one down here, but I forgot
to remove it. UpdateMegamorphicCache could allocate, so the upper scope is
invalid.

On 2013/03/05 17:32:59, Jakob wrote:
> This AssertNoAllocation (and its scope) seem unnecessary, since you have one
on
> line 1035 too.

https://chromiumcodereview.appspot.com/12451003/diff/8016/src/stub-cache.cc
File src/stub-cache.cc (right):

https://chromiumcodereview.appspot.com/12451003/diff/8016/src/stub-cache.cc#n...
src/stub-cache.cc:907: Code::StubType type = handlers->length() == 1
On 2013/03/05 17:32:59, Jakob wrote:
> nit: I have a weak preference for the following formatting, but you can also
> keep what you have if you prefer:
> 
>   Code::StubType type = handlers->length() == 1 ? handlers->at(0)->type()
>                                                 : Code::NORMAL;

Done.

[Toon Verwaest, 2013-03-05 17:38:46]

Committed patchset #2 manually as r13831 (presubmit successful).